Project Summary Understanding the molecular basis of cellular identity in the human brain is a major goal of the BRAIN Initiative and essential for clarifying the cellular origins of diverse brain disorders. This project will use novel, ?top-down? analytical strategies that complement single-cell and single-nucleus methods to define the core transcriptional identities of major cell types in the adult human brain while obviating the need to purify or isolate cells. Our central hypothesis is that covariation between the abundance of cell types and transcripts can be estimated through integrative gene coexpression analysis of intact tissue samples, and this information can be used to construct quantitative cell type definitions, perform mathematical modeling of gene expression, and identify cell type-specific transcriptional differences between biological systems. In Aim 1, we will integrate cell type- specific gene coexpression modules from >60 datasets and >7000 neurotypical adult human brain samples to determine consensus transcriptional profiles of major cell types. These profiles will suggest which major cell types primarily express genes that have been implicated in human brain disorders, identify novel biomarkers, and help to establish the 'ground truth' for assessing the validity of human cell types derived in vitro for disease modeling. We will also leverage highly recurrent gene coexpression relationships to estimate cellular abun- dance and develop predictive models of gene expression in adult human brain samples. These models will im- prove reproducibility through concrete predictions that can be tested in new human brain transcriptomes (in- cluding those from pathological samples) as they become available; they will also lead to new analytical strate- gies that go beyond differential expression analysis to reveal subtle transcriptional perturbations associated with pathology. In Aim 2, we will replicate the goals of Aim 1 in mice and implement a comprehensive effort to identify binary (on/off) expression differences in major cell types between human and mouse brains. In Aim 3, we will assess the extent of regional variation in the transcriptional identities of major CNS cell types in the adult human brain. Expected outcomes include consensus transcriptional profiles of astrocytes, oligodendro- cytes, microglia, neurons, ependymal cells, and endothelial cells; rigorous mathematical models that can accu- rately predict expression levels for thousands of genes in de novo human brain transcriptomes; and new tools and reagents for studying CNS cell types and subtypes. This project is innovative because it challenges the status quo that cells must be physically isolated to study their molecular properties; it also introduces a novel concept and metric called gene expression fidelity for defining cellular identity. Our studies will have a positive impact by providing an unprecedented resource for identifying transcriptional processes that distinguish cell types among human brain regions, species, and disease states, and will contribute directly to our long-term goal of constructing a comprehensive cellular taxonomy of the human CNS from molecular data.